Guidance for Industry Vaginal Microbicides: Development for the

Prevention of HIV Infection

DRAFT GUIDANCE

This guidance document is being distributed for comment purposes only.

Comments and suggestions regarding this draft document should be submitted within 90 days of publication in the Federal Register of the notice announcing the availability of the draft guidance. Submit electronic comments to http://www.regulations.gov. Submit written comments to the Division of Dockets Management (HFA-305), Food and Drug Administration, 5630 Fishers Lane, rm. 1061, Rockville, MD 20852. All comments should be identified with the docket number listed in the notice of availability that publishes in the Federal Register.

For questions regarding this draft document contact Charu Mullick, M.D. or Debra Birnkrant, M.D. at 301-796-1500.

U.S. Department of Health and Human Services Food and Drug Administration Center for Drug Evaluation and Research (CDER)

November 2012 Clinical Antimicrobial

I:\10476dft.doc 11/07/12

Guidance for Industry Vaginal Microbicides: Development for the

Prevention of HIV Infection

Additional copies are available from:

Office of Communications, Division of Drug Information Center for Drug Evaluation and Research Food and Drug Administration 10903 New Hampshire Ave., Bldg. 51, rm. 2201 Silver Spring, MD 20993-0002 Tel: 301-796-3400; Fax: 301-847-8714; E-mail: [email protected] http://www.fda.gov/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/default.htm

U.S. Department of Health and Human Services Food and Drug Administration Center for Drug Evaluation and Research (CDER)

November 2012 Clinical Antimicrobial

TABLE OF CONTENTS

I. INTRODUCTION...... 1 II. BACKGROUND ...... 2 III. DEVELOPMENT PROGRAM ...... 3 A. General Considerations...... 3 1. Nonclinical Considerations ...... 3 a. Nonclinical safety ...... 3 b. Nonclinical virology...... 4 c. Other nonclinical studies ...... 7 2. Drug Development Population ...... 8 3. Early Phase Clinical Considerations...... 8 a. Safety and pharmacokinetic considerations...... 8 b. Drug product characteristics that effect end-user acceptability...... 9 c. Dose selection...... 9 4. Efficacy Considerations...... 9 5. Safety Considerations ...... 11 a. Adequacy of the safety database...... 11 b. General safety considerations...... 11 c. Safety in specific populations...... 12 d. Other safety considerations ...... 15 B. Specific Considerations for Effectiveness Trials ...... 16 1. Trial Design, Randomization, and Endpoints...... 16 2. Choice of Comparator ...... 16 3. Enrollment Criteria...... 17 4. Trial Procedures ...... 18 5. Trial Duration...... 18 6. Statistical Considerations ...... 18 a. Endpoint analysis...... 18 b. Strength of evidence ...... 19 c. Adherence...... 19 d. Interim analysis and data monitoring committee...... 20 e. Missing data and sensitivity analysis...... 20 7. Combination Products ...... 20 8. Risk-Benefit Considerations...... 22 C. Other Considerations...... 22 1. Clinical Virology...... 22 2. Additional Clinical Pharmacology ...... 23 3. Chemistry, Manufacturing, and Controls ...... 23 4. Condom/Device Function Studies...... 24 5. Labeling Considerations...... 24 REFERENCES...... 25

Contains Nonbinding Recommendations Draft — Not for Implementation 1 Guidance for Industry1 2 Vaginal Microbicides: Development for the 3 Prevention of HIV Infection 4 5 6 7 8 This draft guidance, when finalized, will represent the Food and Drug Administration’s (FDA’s) 9 current thinking on this topic. It does not create or confer any rights for or on any person and 10 does not operate to bind FDA or the public. You can use an alternative approach if the approach 11 satisfies the requirements of the applicable statutes and regulations. If you want to discuss an 12 alternative approach, contact the FDA staff responsible for implementing this guidance. If you 13 cannot identify the appropriate FDA staff, call the appropriate number listed on the title page of 14 this guidance. 15 16 17 18 19 I. INTRODUCTION 20 21 This guidance provides recommendations for the development of vaginal microbicides regulated 22 within the Center for Drug Evaluation and Research (CDER) at the Food and Drug 23 Administration (FDA) for the prevention of human immunodeficiency virus (HIV) infection. 24 Specifically, this guidance addresses the FDA’s current thinking regarding the overall 25 development program and clinical trial designs to support the development of vaginal 26 microbicide drug products.2 This draft guidance is intended to serve as a focus for continued 27 discussions among review divisions, pharmaceutical sponsors, the academic community, and the 28 public.3 29 30 Information in this guidance is also generally relevant for developing vaginal microbicides that 31 are part of a drug-device combination product. Guidance on development and testing of devices 32 can be obtained from the Center for Devices and Radiological Health (CDRH). 33 34 For the purposes of this guidance, we define vaginal microbicides as intravaginal drug products 35 that reduce the risk of HIV acquisition. Vaginal microbicides are designed to be self- 36 administered products. Microbicides can be developed as vaginal formulations including gels,

1 This guidance has been prepared by the Division of Antiviral Products in the Center for Drug Evaluation and Research (CDER) in cooperation with the Division of Reproductive and Urologic Products in CDER at the Food and Drug Administration.

2 For the purposes of this guidance, all references to drugs include both human drugs and therapeutic biological products unless otherwise specified.

3 In addition to consulting guidances, sponsors are encouraged to contact the divisions to discuss specific issues that arise during the development of vaginal microbicide drug products.

1 Contains Nonbinding Recommendations Draft — Not for Implementation 37 creams, tablets, and drug-impregnated vaginal rings. Microbicides therefore offer a women- 38 initiated HIV prevention method, which would form a useful addition to the existing prevention 39 interventions. Sponsors can choose to develop products with coitally dependent dosing (pre- 40 coital dosing, or post-coital dosing, or pre-coital plus post-coital dosing) or a coitally 41 independent dosing scheme (e.g., daily dosing, intermittent dosing, or sustained release 42 formulation such as vaginal ring). 43 44 This guidance does not address other forms of HIV prevention such as prophylactic vaccination 45 or prevention mediated exclusively by physical barrier devices (e.g., male condoms). Inquiries 46 regarding vaccines should be addressed to the Center for Biologics Evaluation and Research. 47 Inquiries regarding physical barrier devices should be addressed to CDRH. Prevention of 48 sexually transmitted infections (STI) other than HIV also is not addressed in this guidance. 49 Additionally, general issues of clinical trial design or statistical analyses are not addressed in this 50 guidance. Those topics are addressed in the ICH guidances for industry E9 Statistical Principles 51 for Clinical Trials and E10 Choice of Control Group and Related Issues in Clinical Trials.4 52 53 Sponsors considering development of vaginal microbicides are encouraged to consult this 54 guidance and to communicate with the FDA through the pre-investigational new drug application 55 consultation program and throughout drug development. 56 57 FDA’s guidance documents, including this guidance, do not establish legally enforceable 58 responsibilities. Instead, guidances describe the Agency’s current thinking on a topic and should 59 be viewed only as recommendations, unless specific regulatory or statutory requirements are 60 cited. The use of the word should in Agency guidances means that something is suggested or 61 recommended, but not required. 62 63 64 II. BACKGROUND 65 66 The natural history of HIV infection includes a brief symptomatic period characterized by 67 intense viral replication, or acute HIV infection. This acute phase is followed by a clinically 68 latent period and eventual progression to a state of profound immunodeficiency known as 69 acquired immunodeficiency syndrome (AIDS). 70 71 Sexual transmission accounts for the majority of HIV infections both in the United States and 72 globally. Behavior change through counseling, male and female condoms, antiretroviral therapy 73 for the infected partner, voluntary male medical circumcision, and treatment of STIs can reduce 74 the risk of HIV acquisition. However, despite these prevention methods, HIV incidence in the 75 United States has not declined and remains stable at 56,300 new infections annually (Hall, Song, 76 et al. 2008). Worldwide, the annual incidence is estimated to be about 2.7 million infections 77 (UNAIDS 2011). 78

4 We update guidances periodically. To make sure you have the most recent version of a guidance, check the FDA Drugs guidance Web page at http://www.fda.gov/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/default.htm.

2 Contains Nonbinding Recommendations Draft — Not for Implementation 79 In 2012 the first drug product for oral pre-exposure prophylaxis (PrEP), tenofovir/emtricitabine, 80 was approved to reduce the risk of sexually acquired HIV-infection in adults at high risk. Data 81 from trials in both men who have sex with men and heterosexual populations supported this 82 approval. Despite its efficacy, this intervention poses several challenges including adherence to 83 daily pill intake, the potential for renal and bone toxicities, the potential for decreased condom 84 use, and the development of resistance to drugs that are also used for HIV treatment. 85 86 With respect to women and the HIV epidemic, present trends indicate that women account for 23 87 percent of new infections in the United States (CDC 2010) and 50 percent of all new infections 88 globally. Heterosexual transmission plays a major role in HIV infection in women, and among 89 the available options for reducing heterosexual male-to-female transmission, virtually all require 90 male partner agreement to be effective. For example, using male condoms, which are widely 91 available through condom promotion programs, depends on the male partner’s level of condom 92 acceptance (Kulczycki, Kim, et al. 2004). In many settings, women are unable to insist on or 93 negotiate use of condoms or another available prevention method. In these situations, gender 94 inequality, reliance on men for economic security, and compromised relationship dynamics may 95 contribute to the prevention challenge (Aziz and Smith 2011). Taken together, these limitations 96 emphasize the need for novel prevention approaches that allow women to independently control 97 their HIV acquisition risk. Lastly, a woman’s product preference may change over time 98 depending on the desire to conceive, partner’s product preference, perceived risk for other STIs, 99 dosing convenience, or another consideration. At-risk women need multiple prevention options 100 to meet their needs. In this context, vaginal microbicides offer the potential of a unique female- 101 controlled method. 102 103 104 III. DEVELOPMENT PROGRAM 105 106 A. General Considerations 107 108 1. Nonclinical Considerations 109 110 a. Nonclinical safety 111 112 General recommendations for supportive nonclinical safety studies including their design and 113 timing are addressed in other FDA and ICH guidances for industry, such as the ICH guidance for 114 industry M3(R2) Nonclinical Safety Studies for the Conduct of Human Clinical Trials and 115 Marketing Authorization for Pharmaceuticals. Nonclinical considerations specific to vaginal 116 microbicide development are discussed in this guidance. 117 118 A candidate vaginal microbicide should be assessed for the potential to cause inflammation, 119 disruption, erythema, or edema of cervicovaginal epithelium. Local safety and tolerability can 120 be evaluated alone or as part of a repeat-dose toxicology study. Animal toxicity studies should 121 include a scoring system to evaluate the extent of erythema, edema, leukocyte infiltration, and 122 epithelial layer ulceration or disruption. Studies should also incorporate detailed 123 histopathological assessments (i.e., by dividing the vagina into proximal, mid, and distal sections 124 from the application site). For drug products containing an entity originally approved as a

3 Contains Nonbinding Recommendations Draft — Not for Implementation 125 nonvaginal formulation, bridging toxicology and vaginal irritation studies should be conducted. 126 For drug products currently approved as a vaginal formulation for another indication, additional 127 toxicology studies usually are not needed unless the approved vaginal drug product is 128 reformulated. Additional local tolerance studies via rectal application should be completed 129 before evaluating the rectal safety profile in humans. 130 131 Systemic drug absorption following vaginal administration should be evaluated because the 132 extent of plasma exposures is a key determinant of systemic toxicity. When lower systemic 133 exposures are achieved following vaginal administration in animals compared to humans, 134 nonclinical studies using an alternate route of administration may be needed to attain the 135 appropriate exposure. 136 137 At least one carcinogenicity study using the vaginal route of administration should be conducted 138 to evaluate local tumorigenic potential of the vaginal microbicide. For drug products containing 139 entities previously evaluated in oral carcinogenicity studies, one topical carcinogenicity study 140 may be needed. For detailed information for animal carcinogenicity studies, sponsors should 141 refer to the ICH guidances for industry S1A The Need for Long-term Rodent Carcinogenicity 142 Studies of Pharmaceuticals, S1B Testing for Carcinogenicity of Pharmaceuticals, and S1C(R2) 143 Dose Selection for Carcinogenicity Studies. 144 145 Reproductive toxicology studies should be completed and submitted for FDA review before 146 initiating the first-in-human study. At least one embryo/fetal development study should be 147 conducted with the vaginal route of administration. Systemic exposures should be evaluated in 148 nonclinical reproductive toxicology studies. Similar to repeat-dose toxicology studies discussed 149 previously, if animal exposures are lower than human exposures, then sponsors should conduct 150 animal studies with other routes of administration that yield relevant systemic exposures. 151 152 b. Nonclinical virology 153 154 The biology of HIV vaginal transmission is not fully understood and the following 155 recommendations may be modified as the field evolves. In addition to identifying the 156 mechanism of action, nonclinical virology studies should address the following: 157 158  Quantification of antiviral activity and the potential for cellular toxicity 159  Selection and characterization of resistant HIV 160  Effects on other sexually transmitted pathogens and the assays used to detect them 161  Effects on normal vaginal microflora 162 163 Validated animal models for predicting efficacy in humans do not exist. Sponsors often use 164 animal models to obtain supportive activity data; however, they are not needed to support 165 approval of an indication. Generally, animal model studies are designed to show a protective 166 effect against vaginal challenge with simian immunodeficiency virus or chimeric simian/human 167 immunodeficiency virus in nonhuman primates. Humanized mouse models might also prove to 168 be useful in evaluating activity against vaginal challenge of HIV; however, the limited 169 availability of human hematopoietic progenitor cells may restrict the use of these models 170 (Denton, Estes, et al. 2008; Berges, Akkina, et al. 2008).

4 Contains Nonbinding Recommendations Draft — Not for Implementation 171 172 Quantification of antiviral activity and cellular toxicity 173 174 Because the principal objective of the microbicide is to prevent HIV transmission, studies should 175 demonstrate inhibition of HIV replication in cell culture. Dose-response curves should be 176 generated to determine the range of antiviral activity. These results should reflect the 177 concentration of the drug required to reduce HIV replication by 50 percent (i.e., 50 percent 178 effective concentration (EC50) value). 179 180 Well-characterized laboratory strains of HIV-1 should be used in initial evaluations to validate 181 antiviral activity. After demonstrating antiviral activity against standard laboratory strains, 182 evaluation should be expanded to cover a broad range of clinically relevant viruses including 183 those isolated from the reproductive tract, multiple isolates representing each of the HIV-1 184 clades, CCR5 and dual CCR5/CXCR4 co-receptor tropic strains, and HIV-2. The candidate 185 microbicide should also have demonstrable antiviral activity against several (greater than or 186 equal to 20) temporally and geographically distinct isolates, including U.S. strains and strains 187 endemic to regions where clinical trials will be conducted. Additionally, antiviral activity should 188 be assessed against multiple isolates from each HIV-1 clade and HIV-2. The median and range 189 of EC50 values should be provided. 190 191 Sponsors should assess antiviral activity in peripheral blood mononuclear cells, primary 192 macrophage and dendritic cell cultures, and cervicovaginal explants because these cultures 193 represent the cell types likely involved in sexual transmission of HIV (Wu and Kewalramani 194 2006). Assays including continuous cell lines such as ME-180 cervical epithelial cells (CD4- 195 transformed cell line) or GHOST X4/R5 human osteosarcoma cells may provide supporting data; 196 however, these cannot be substituted for primary cultures because their biological relevance to 197 human infection is unclear at this time. Primary cultures should be harvested from several 198 different donors to verify antiviral activity across a genetically diverse population of subjects. 199 These studies are particularly important for drugs targeting host proteins (e.g., CCR5 or CD4) 200 because polymorphisms may affect antiviral activity. 201 202 In addition to the diversity of viral variants and cellular targets, other variables associated with 203 transmission may affect activity. Assays should be performed under conditions consistent with 204 drug product use to determine: 205 206  Antiviral activity across a range of multiplicities of infections 207  Antiviral activity following pH transition from an approximate pH 4 to pH 7 208  Protection against infection by cell-free and cell-associated HIV-1 virus 209  Antiviral activity in the presence of seminal plasma and cervicovaginal lavage fluid 210  Virucidal kinetics (in seconds or minutes) for nonspecific agents inactivating virus 211 212 Furthermore, sponsors should verify antiviral activity of the proposed microbicide formulation to 213 identify excipient effects. The results for the effect of each variable should include the fold-shift 214 in EC50 value relative to infection under standard conditions. 215

5 Contains Nonbinding Recommendations Draft — Not for Implementation 216 Studies determining range of antiviral activity should be conducted under biologically relevant 217 conditions and results should demonstrate that antiviral activity is not a result of damage to host 218 cells. Cytotoxic compounds may reduce capacity of the host cell to support HIV replication that 219 may be misinterpreted as direct antiviral activity. To differentiate direct antiviral activity from 220 negative effects on the host cell, cytotoxicity should be quantified as the cytotoxicity 221 concentration (CC50) value, the drug concentration required to reduce culture metabolism or 222 viability by 50 percent. 223 224 Additionally, the therapeutic index (TI) should be calculated as the ratio of the CC50 and EC50 225 values (TI = CC50/EC50). Generally, a higher TI value indicates a more specific antiviral effect 226 and greater likelihood that an effective concentration of the drug can be achieved without 227 undesired off-target effects on the host cell. Further, the TI can be useful in comparing different 228 candidate microbicides or in quantifying relative activity of a particular microbicide against 229 different viral variants or different cell types. In general, we do not recommend development of 230 a drug with a TI less than or equal to 10; sponsors interested in pursuing development of a drug 231 with TI less than or equal to 10 should discuss the basis for pursuing development with the FDA. 232 Sponsors should refer to the National Institute of Allergy and Infectious Diseases/Division of 233 Acquired Immunodeficiency Syndrome (DAIDS)/National Institutes of Health nonclinical 234 resources for assistance in developing their microbicide.5 235 236 Selection and characterization of resistant virus 237 238 HIV seroconversion during microbicide use may result in resistant virus and affect HIV 239 treatment options. Understanding the pathway to resistance development (i.e., the amino acid 240 substitutions that confer reduced susceptibility to an antiviral drug) is useful in understanding the 241 potential risks of microbicide failure and in guiding resistance monitoring plans in clinical trials. 242 Characterization of resistant isolates also can provide supporting evidence for the proposed 243 mechanism of action. 244 245 Because the genetic barrier to resistance may vary as a function of drug concentration, resistant 246 variants should be selected in cell-culture at different microbicide concentrations. Several 247 independent resistant isolates should undergo genotypic and phenotypic testing. Resistance- 248 associated substitutions should be verified by phenotypic characterization of site-directed 249 recombinant viruses expressing the mutant protein. If substitutions arise in viral proteins or 250 protein complexes targeted by approved antiretroviral drugs, cross-resistance should be assessed 251 including: 252 253  Assessment of the susceptibility of approved drugs to microbicide-resistant variants 254  Assessment of susceptibility of the microbicide to variants resistant to approved 255 antiretroviral drugs 256

5 See http://www.niaid.nih.gov/labsandresources/resources/atrg/pages/microbicidesprevhivtrans.aspx.

6 Contains Nonbinding Recommendations Draft — Not for Implementation 257 Effect on STIs and normal human microflora 258 259 Local pathologic changes associated with STIs as well as vaginal microflora alterations may 260 affect the risk for vaginal HIV transmission. Sponsors should test for antimicrobial activity 261 against common STI pathogens, such as Neisseria gonorrhoeae, Chlamydia trachomatis, herpes 262 simplex virus type-2, and trichomonas. Testing antimicrobial activity is of particular importance 263 for drug products with a nonspecific mechanism of action and/or a low TI. Testing also should 264 be performed to assess static and cidal activity on normal resident microflora such as 265 Lactobacilli. Sponsors should be aware that the microbicide may contain components inhibitory 266 to STI assays (e.g., sulfated polysaccharides can affect certain polymerase chain reaction assays) 267 that may interfere with STI diagnosis. Hence, microbicide effects on sensitivity of STI 268 diagnostic assays used in clinical trials should be evaluated using concentrations consistent with 269 drug product use. 270 271 c. Other nonclinical studies 272 273 Condom compatibility studies are nonclinical studies intended to evaluate effects of a 274 microbicide on the physical properties of condoms. Concomitant use of microbicide and 275 condoms is expected both in clinical trials as well as real-world use following microbicide 276 approval. Condom compatibility studies are needed to determine whether use of microbicide 277 with condoms affects the rate of condom failure compared to condoms used alone. 278 279 Sponsors should refer to the American Society for Testing and Materials (ASTM) D7661-10, 280 Standard Test Method for Determining Compatibility of Personal Lubricants with Natural 281 Rubber Latex Condoms, for study design and methodology. Studies should include both male 282 and female condoms composed of a variety of materials including natural rubber latex, 283 polyisoprene, polyurethane, and nitrile. In addition to baseline, conditioned, and positive 284 controls, concurrent evaluation of the effects of microbicide and placebo on each condom type 285 should be performed. Data analyses and presentation should follow recommendations outlined 286 in the above-mentioned ASTM standard. Viral penetration testing assessing drug effects on the 287 barrier properties of male and female condoms also should be performed. Sponsors are strongly 288 encouraged to provide study proposals for FDA review and comment before initiating 289 nonclinical compatibility studies. 290 291 Positive or equivocal findings in nonclinical compatibility studies may prompt the need for 292 clinical data to fully understand effects of microbicide on condom function. Clinical data 293 obtained from condom function studies are discussed further in section III.C.4., Condom 294 Device/Function Studies. 295 296 Sponsors should also evaluate the effect of a microbicide on the physical properties of other 297 commonly used barrier contraceptive devices (e.g., diaphragms). Before initiating these studies, 298 sponsors should contact the FDA to obtain feedback for their proposed plan and/or protocol. 299

7 Contains Nonbinding Recommendations Draft — Not for Implementation 300 2. Drug Development Population 301 302 Clinical development should be pursued in women at risk of acquiring HIV through sexual 303 transmission. Microbicide effectiveness trials generally are conducted in high HIV prevalence 304 areas or populations because the HIV seroconversion rate is a critical factor in determining 305 sample size. Because the largest number of new HIV infections occurs outside the United States, 306 clinical effectiveness data supporting an indication will likely be obtained from trials conducted 307 at foreign sites. FDA regulations permit accepting foreign trial data in support of a marketing 308 application, provided the foreign studies meet the same requirements in 21 CFR part 312 as are 309 applicable to U.S. studies conducted under an investigational new drug application.6 310 Additionally, the clinical development program should include U.S. subjects to ensure 311 applicability of data to the U.S. population. 312 313 3. Early Phase Clinical Considerations 314 315 Because HIV seroconversion is an infrequent event even in high prevalence areas, proof of 316 concept can be measured only in trials with relatively large sample sizes. As a result, vaginal 317 microbicide development generally proceeds directly from phase 1 into large-scale phase 2b or 318 phase 3 trials. The objective of early phase clinical development is to provide sufficient data for 319 preliminary safety, tolerability, acceptability, and pharmacokinetics to support rational drug 320 development and choice of dose or doses to take forth into late-phase trials. 321 322 a. Safety and pharmacokinetic considerations 323 324 Initial safety should be evaluated in sexually active healthy women at low risk for potential 325 confounders for evaluating safety/tolerability, such as STIs or baseline cervicovaginal 326 abnormalities. Safety evaluations should focus on local toxicities of the cervicovaginal area and 327 the female reproductive tract, as well as systemic toxicities. Initial studies can obtain safety data 328 reflecting daily drug product use for at least one complete ovulatary/menstrual cycle. Studies of 329 at least 3 to 6 months dosing duration are needed to identify toxicities from cumulative drug 330 product use. Generally, data from at least 100 to 200 subjects are needed to characterize the 331 preliminary safety profile. 332 333 Systemic absorption should be determined early in drug development because of important 334 safety and drug development implications. For systemically absorbed drug products, the 335 potential for systemic adverse events should be monitored from early trials onward. Knowledge 336 about systemic absorption influences the type of clinical pharmacology studies needed for drug 337 product development. For example, typical clinical pharmacology studies (e.g., systemic drug 338 interaction studies, renal impairment) may not be needed if plasma exposure is low. For drug 339 products containing either a drug originally approved for HIV treatment or a new molecular 340 entity in the same class as an approved drug for HIV treatment, the presence of sub-inhibitory 341 but detectable systemic drug concentrations can potentially result in resistance or cross- 342 resistance, respectively, in the event of HIV seroconversion. 343

6 See the guidance for industry Acceptance of Foreign Clinical Studies.

8 Contains Nonbinding Recommendations Draft — Not for Implementation 344 Local pharmacokinetic (PK) evaluations should include serial assessment of cervicovaginal fluid 345 concentrations at different time points after dosing to estimate the level of local exposure as well 346 as to determine elimination of the drug locally. Local drug distribution can be quantified by 347 collection of cervicovaginal fluid samples from multiple regions of the vaginal lumen. 348 Researchers have used magnetic resonance imaging and other imaging techniques as another 349 mechanism to determine local and regional drug distribution. Hypothetically, to be optimally 350 effective, a vaginal microbicide should uniformly coat the entire vaginal lumen and cervix. 351 However, there is presently no clear evidence linking extent of local coverage or tissue 352 distribution or local drug exposure to clinical outcome for vaginal microbicides. Tissue samples 353 obtained by biopsy, when medically necessary, can be used to quantify the extent to which the 354 drug is taken up by cervicovaginal epithelium. In addition, local PK data collected in early 355 studies should include an assessment of microbicide drug concentrations during menses, both 356 with and without tampon use. For vaginal ring microbicides, sponsors should determine local 357 tissue drug concentrations at different time points after ring removal to provide an estimate of the 358 elimination of the drug locally following ring expulsion or loss. 359 360 b. Drug product characteristics that effect end-user acceptability 361 362 Phase 1 trials should assess drug product characteristics that affect user acceptability early in 363 drug development. In some instances, drug product reformulation may be necessary to achieve 364 maximum acceptability before embarking on large scale trials to avoid poor adherence and loss 365 of potential effectiveness. Poor or low acceptability may be secondary to unfavorable physical 366 attributes (e.g., viscosity, odor, color, taste). For semisolid gel formulations, vaginal retention 367 should be assessed because excessive vaginal leakage after application may be undesirable and 368 also lead to poor distribution and possibly compromised drug effect. 369 370 c. Dose selection 371 372 Dose selection poses a unique challenge to microbicide development. Typical dose-response or 373 exposure-response analyses are unlikely to be helpful for microbicides unless validated surrogate 374 markers of protection are identified. In vitro antiviral activity data combined with local exposure 375 data (e.g., vaginal and cervical concentrations) can be used to predict a minimally efficacious 376 dose. Protein-binding of the drug product in vaginal fluid should be taken into consideration for 377 highly protein-bound drug products when estimating target concentrations. Additionally, animal 378 toxicity findings and safety findings in phase 1 clinical trials can provide further information 379 guiding human dose selection. Another consideration is systemic absorption; in some instances 380 higher doses may not be preferred because of increased systemic absorption and a greater 381 likelihood of systemic adverse events. 382 383 4. Efficacy Considerations 384 385 Despite advances in the HIV and microbicide field, there are several challenges in the clinical 386 development of vaginal microbicides. Some challenges in evaluating clinical efficacy are 387 addressed below. 388

9 Contains Nonbinding Recommendations Draft — Not for Implementation 389 In addition to biological efficacy, variables such as drug product adherence, concurrent use of 390 other prevention methods (e.g., condoms, oral PrEP), and frequency of high-risk sexual behavior 391 are closely linked to the overall effectiveness of a vaginal microbicide. Because usage rates of 392 microbicide and/or other prevention methods as well as sexual behavior patterns are expected to 393 fluctuate over time, and long-term safety events (e.g., epithelial disruptions) may potentially 394 reduce efficacy over time, longer duration effectiveness trials are preferred because they can 395 capture the effect of these variables and are more likely to reflect the real-world effects (Lagakos 396 and Gale 2008). 397 398 Effectiveness trials should measure the rate of new HIV infections. Given the relatively low 399 incidence of seroconversion even in high-prevalence populations, a large sample size usually is 400 necessary to provide adequate power to detect a statistically significant effect on HIV 401 seroincidence. As with other drug trials, sample size calculations should account for the 402 anticipated effect size of the drug product, as well as loss to follow-up and dropouts caused by 403 adverse events. An additional consideration is the anticipated pregnancy rate, because pregnant 404 women in microbicide trials should be taken off the drug product unless specific criteria are met 405 (see section III.A.5.c., Safety in specific populations (Pregnant women)). 406 407 Risk-reduction counseling and promoting condom use are ethical imperatives during the conduct 408 of microbicide trials. Additionally, an approved oral PrEP agent can be offered in the trial as 409 part of the background prevention package. Offering oral PrEP depends on several 410 considerations including acceptability as standard HIV prevention locally and implementation in 411 regions where trials are conducted. Alternatively, trials can be designed to enroll subjects who 412 refuse oral PrEP as a result of intolerance, side effects, or personal preference. Importantly, 413 condom promotion and other prevention interventions are likely to reduce the infection rate in 414 the trial and further increase the trial sample size. 415 416 Accelerated approval (21 CFR part 314, subpart H), based on a surrogate endpoint considered 417 reasonably likely to predict clinical benefit for a serious or life-threatening disease, is not 418 applicable for an HIV prevention indication. The endpoint of HIV infection, although a 419 laboratory measurement, is considered to be reliably predictive of progressive clinical disease. 420 At this time, other surrogates that are predictive of laboratory infection of HIV have not been 421 defined. The assay used to diagnose HIV infection should be the most reliable assay available at 422 the time of the trial. 423 424 Sponsors can request that the FDA expedite drug product development through mechanisms such 425 as priority review and fast track designation.7 The fast track designation allows for frequent 426 interactions with the FDA and permits rolling review when a new drug application (NDA) is 427 submitted. Proposals for fast track designation can be considered at any time during 428 development depending on appropriate fulfillment of the designated criteria.8 429

7 See http://www.fda.gov/ForConsumers/ByAudience/ForPatientAdvocates/SpeedingAccesstoImportantNewTherapies.

8 See the guidance for industry Fast Track Drug Development Programs – Designation, Development, and Application Review.

10 Contains Nonbinding Recommendations Draft — Not for Implementation 430 5. Safety Considerations 431 432 a. Adequacy of the safety database 433 434 The safety database should consist of approximately 3,000 subjects exposed to the proposed or 435 to-be-marketed dose of a vaginal microbicide for at least 12 months. A proportion of subjects 436 should be followed for longer duration as outlined in section III.B.5., Trial Duration. Data from 437 more subjects may be needed if safety concerns are identified during development. 438 Alternatively, a smaller safety database may be considered adequate if the drug product is 439 already approved for use by another route/formulation or is unapproved but has been evaluated 440 substantially using other routes of administration. Sponsors are strongly encouraged to discuss 441 their proposed safety database with the FDA at an end-of-phase 2 meeting or earlier. 442 443 b. General safety considerations 444 445 Local vaginal and cervical safety is a critical consideration for a vaginal microbicide. Local 446 toxicity should be assessed through symptoms and signs for genitourinary and reproductive 447 adverse events and by pelvic examination. Evaluations should focus on symptoms and signs 448 representing genital irritation, inflammation, or mucosal breakdown. Pelvic examination should 449 include visual inspection and speculum examination. Comparative safety data from randomized, 450 double-blind trials using an appropriate placebo control are preferred to allow clear interpretation 451 of safety findings. Severity grading of cervicovaginal abnormalities should be based on accepted 452 grading criteria for genital toxicity such as the National Institutes of Health DAIDS genital 453 toxicity table.9 454 455 Colposcopy should be performed in at least one multiple-dose phase 1 trial in sexually active 456 women conducted early in drug product development. This technique was originally developed 457 to detect local malignancy; specifically for microbicide trials, the examination should focus on 458 findings reflecting potential drug toxicities including epithelial disruption. Sponsors should refer 459 to standard criteria for colposcopy technique and training for vaginal microbicides.10 Depending 460 on the safety profile and colposcopic findings in phase 1, we will determine the need for 461 colposcopy in subsequent phase trials. We do not consider vaginal biopsies necessary unless 462 indicated by local toxicity findings. Trials should also monitor for adverse events reflecting drug 463 effects on the uterus, fallopian tubes, and ovaries that may arise from regional drug distribution. 464 465 Sponsors should perform assessments for microbicide effects on vaginal pH, balance of vaginal 466 microflora, and the frequency of other STIs. Significant shifts in local microflora may have 467 clinical implications because the normal vaginal microflora is thought to play a role in 468 preventing HIV-1 infection and other STIs (Myer, Kuhn, et al. 2005). Certain type of microflora 469 imbalance or decreases in particular flora species can also increase the likelihood of bacterial 470 vaginosis, urinary tract infections including urosepsis, and pelvic infections. 471

9 See http://rsc.tech-res.com/safetyandpharmacovigilance/gradingtables.aspx.

10 See the World Health Organization Manual for the Standardization of Colposcopy for the Evaluation of Vaginal Products at http://www.who.int/reproductivehealth/publications/rtis/RHR_04.2/en/index.html.

11 Contains Nonbinding Recommendations Draft — Not for Implementation 472 Systemic adverse reactions may arise if a drug product is systemically absorbed following 473 vaginal administration. At a minimum, systemic safety should be evaluated through adverse 474 event assessment and routine laboratory tests such as hematology and chemistry parameters. The 475 need for additional evaluation depends on the level of systemic absorption and expected or 476 known risks of the drug product as indicated by nonclinical findings or toxicities observed in 477 earlier human trials. For a microbicide originally approved as an oral formulation, sponsors 478 should consider the established safety profile when forming a targeted safety assessment plan for 479 trial protocols. For example, if thyroid function abnormality is a recognized toxicity with the 480 oral formulation, appropriate laboratory testing of thyroid function should be included if 481 significant systemic absorption is expected. Grading of nongenital adverse events and laboratory 482 abnormalities should follow commonly used and accepted toxicity grading schemes. 483 484 For drug products intended for vaginal retention for a specified time period (e.g., a 28-day 485 intravaginal ring), sponsors should collect safety data reflecting greater than the prescribed 486 duration of exposure to characterize safety in case of drug product overuse. 487 488 Safety evaluations in phase 1 trials can be performed at weekly intervals or less frequently. In 489 late-stage trials, evaluations should be performed within the first month of starting the drug 490 product and then at least once every 1 or 2 months. The protocol should have provisions to 491 contact subjects or allow unscheduled visits for managing adverse events as needed. All visits 492 should include safer sex and HIV risk reduction counseling and provision of male condoms to all 493 subjects. 494 495 In addition to evaluating genital toxicity in women, sponsors should conduct male tolerance 496 studies to identify penile toxicity. The need for assessing oral safety (following oral exposure) 497 depends on whether the microbicide contains an orally approved antiretroviral, the anorectal 498 safety profile (pertinent because of histological similarities in mucosal lining), and findings in 499 acceptability studies evaluating taste. 500 501 c. Safety in specific populations 502 503 In addition to the general safety considerations discussed previously, this section outlines safety 504 perspectives for specific populations, namely pregnant women and adolescents. 505 506 Pregnant women 507 508 To protect the fetus from research-related risks, microbicide trials have historically excluded 509 pregnant women from enrollment and disallowed women who became pregnant to continue to 510 use the investigational drug product. In the real-world setting, however, an approved 511 microbicide may be used by pregnant women despite lack of data in this special population. We 512 recognize that complete lack of safety data in pregnant women at time of NDA submission is not 513 optimal. Ideally, microbicide safety data in pregnant women should be obtained methodically in 514 a prospective controlled trial setting with careful monitoring of subjects. Because of these 515 considerations, the FDA’s thinking regarding the evaluation of women who become pregnant 516 while participating in microbicide trials is evolving. In addition, both the woman and the fetus 517 may stand to benefit from a microbicide that could potentially prevent HIV infection.

12 Contains Nonbinding Recommendations Draft — Not for Implementation 518 519 In some instances, use of an investigational microbicide may be allowed in women who become 520 pregnant while participating in microbicide clinical trials. The decision to allow use of the drug 521 product in women who become pregnant while participating in microbicide clinical trials 522 depends on the safety and PK profile of the drug product in nonclinical studies and early clinical 523 trials. To assess whether pregnant women should continue drug product use, the following 524 information will be considered in the decision-making process: 525 526  Completed reproductive toxicology studies including data from Segment I, II, and III 527 studies 528 529  Completed genotoxicity studies 530 531  Chronic toxicity studies in two species to support the duration in human trials 532 533  Data on systemic absorption of the microbicide in nonpregnant female subjects 534 535 Additionally, the microbicide trial design should have the following provisions for women who 536 become pregnant. 537 538  Women who become pregnant during the trial should be re-consented. Information about 539 potential risk to the fetus following microbicide exposure should be included in the 540 original informed consent and discussed again with the pregnant subject. Women who 541 choose to discontinue the investigational drug product but remain in the trial should be 542 followed for pregnancy and fetal outcome data. 543 544  Pregnant women who choose to continue in the trial should undergo increased safety 545 monitoring including more frequent visits, laboratory testing, and fetal monitoring. In 546 addition, the protocol should include a toxicity monitoring and safety management plan 547 for pregnant women and their fetuses. Protocol safety monitoring should take into 548 account current standards of antepartum care in the trial countries. 549 550  PK data should be collected in a subset of pregnant women during pregnancy, in the 551 postpartum period, and during lactation to characterize changes in systemic exposure 552 related to any alteration in local absorption.11 553 554  The exposed infant should be followed from birth up to 1 year for collection of data 555 during this time frame. Sponsors should obtain FDA concurrence for the type of data 556 considered acceptable. 557 558  Women who become pregnant during the trial should be followed in a pregnancy 559 exposure registry such as the Microbicide Trials Network Registry MTN-016. 560

11 See the draft guidance for industry Pharmacokinetics in Pregnancy — Study Design, Data Analysis, and Impact on Dosing and Labeling. When final, this guidance will represent the FDA’s current thinking on this topic.

13 Contains Nonbinding Recommendations Draft — Not for Implementation 561 The FDA’s decision on a sponsor’s proposal to dose pregnant women will be made on a case-by- 562 case basis. 563 564 Adolescents12 565 566 In the United States, high-risk heterosexual contact is estimated to account for approximately 90 567 percent of transmission in adolescent females aged 13 to 19 years (CDC 2010). In certain 568 countries in Sub-Saharan Africa, HIV prevalence is estimated as high as 20 percent among 569 females aged 15 to 24 years (UNAIDS 2011). Local sociocultural norms and biological factors 570 may contribute to the high infection rate in this age group. Younger women with older sexual 571 partners may be less successful in negotiating male condom use. Cervicovaginal differences 572 between adolescence and adulthood, such as the extent of cervical ectopy, may also play a role. 573 In adolescence, a larger zone of ectopy that is associated with highly vascularized epithelial 574 lining and greater mucosal fragility may increase HIV susceptibility (Moss, Celemetson, et al. 575 1991). 576 577 Conducting clinical trials in adolescents involves important ethical, regulatory, and parental and 578 adolescent consent considerations (Nelson, Lewis, et al. 2010). Generally for drug products 579 approved for adult use, separate clinical studies are needed to establish safety in the pediatric age 580 group, if appropriate. A central consideration in pediatric research is whether the intervention 581 offers the prospect of direct benefit to the enrolled subject. Further, the level of risk associated 582 with the intervention should be considered; an intervention that is not low risk must offer 583 sufficient prospect of direct benefit to justify risk (21 CFR part 50, subpart D). Other influential 584 factors include justification of the intervention in the context of disease severity, comparability 585 of the intervention to available alternatives, and past experience for a drug product or the class. 586 Specifically for vaginal microbicides, key considerations in the decision-making process include 587 the serious and life-threatening nature of HIV disease, high HIV prevalence among adolescents, 588 the likelihood of drug product usage after approval, and the safety profile and efficacy of the 589 drug product. 590 591 After approval for adult use, a vaginal microbicide may be used by individuals younger than 18 592 years, even in the absence of safety data in adolescents. Adolescent safety data should therefore 593 be collected before microbicide approval. Sponsors should consider a two-stage approach 594 whereby initial safety data are collected from older subjects 16 to 18 years of age, followed by 595 recruitment of adolescents younger than 16 years. Sponsors should make every effort to submit 596 safety data from the older adolescent group with the NDA submission. Collecting safety data in 597 the older age group first is appropriate because this reflects the likely target adolescent 598 population. 599 600 Safety data can be collected in a separate adolescent trial or through limited adolescent 601 enrollment in ongoing phase 3 trials. Retention in trials and adherence to the drug product are 602 challenges inherent to this age group that should be taken into consideration. Sexual practices 603 vary by region; therefore, sponsors are strongly encouraged to obtain some of their data from 604 adolescents in the United States. 605

12 For the purposes of this guidance, reference to pediatric subjects includes only the adolescent age group.

14 Contains Nonbinding Recommendations Draft — Not for Implementation 606 d. Other safety considerations 607 608 This section focuses on additional safety considerations such as assessing safety following rectal 609 application of semisolid microbicide preparations, and safety in the postmenopausal age group. 610 611 Rectal safety 612 613 Vaginal and rectal compartments differ in several respects including structural anatomy, local 614 epithelial lining, and the population of target immune cells (Poles, Elliott, et al. 2001). As a 615 result, a microbicide shown to be safe and effective for vaginal use should not be assumed to be 616 safe or effective with rectal administration. Use of an unsafe drug product may be deleterious if 617 it increases HIV susceptibility through irritation or breakdown of the more fragile rectal mucosa. 618 After approval for vaginal use, microbicides may be used by women and men off-label rectally 619 with the intention of preventing infection; therefore, rectal safety data should be available at the 620 time of drug product registration to provide information on any potential significant safety 621 concerns. 622 623 Gastrointestinal (GI) toxicity or anorectal toxicity may be caused by the active ingredient, 624 excipient, or physical attributes of the drug product such as osmolarity. Rectal safety trials 625 should be conducted in healthy adult men and women. Subjects with underlying anorectal or GI 626 conditions including evidence of inflammation at screening anoscopy, history of gastrointestinal 627 bleeding, or untreated rectal STIs should be excluded. Trials should initially evaluate single or 628 multiple rectal doses in subjects who are not planning to engage in anal intercourse for the trial 629 duration. Initial trials should include a minimum dosing duration of approximately 7 to 10 days. 630 Rectal application of lubricants and suppositories or rectal douching should not be allowed. 631 Subjects should be monitored for symptoms and signs of anorectal, GI, and systemic adverse 632 events. Safety laboratory testing should include routine safety parameters such as complete 633 blood count, liver, and renal markers. Lower GI endoscopy should be conducted for visual 634 inspection of mucosal toxicity. 635 636 Sponsors should consider drug product discontinuation in subjects developing GI or abdominal 637 adverse events, and evaluations exploring potential drug toxicity should be undertaken in 638 consultation with a gastroenterology expert. The protocol should include an adequate safety 639 monitoring plan as well as individual and trial stopping criteria. Nonclinical safety findings and 640 the overall clinical safety profile should be used as guides in making the decision for longer 641 duration safety trials. Plasma drug levels should be determined because systemic exposure with 642 rectal administration may not mimic levels attained with vaginal use. 643 644 Randomized, double-blind, and placebo-controlled trials are preferred. Until data supporting 645 placebo effects of hydroxyethyl cellulose (HEC) gel in the rectal compartment are available, 646 safety comparisons with HEC gel should be interpreted with caution. Variations in osmolarity 647 have been shown to affect rectal mucosal integrity (Fuchs, Lee, et al. 2007). Physical attributes 648 such as osmolarity and pH of the placebo gel should match the microbicide gel being evaluated. 649 Applicators considered an acceptable vaginal delivery device may not be suitable for rectal drug 650 product administration and sponsors are encouraged to identify an appropriate delivery device. 651

15 Contains Nonbinding Recommendations Draft — Not for Implementation 652 Safety in postmenopausal women 653 654 Microbicide clinical trials typically enroll sexually active adult women up to the age of 40 or 45 655 years. However, in the United States, the incidence of new HIV infections has risen in women 656 over the age of 50. Postmenopausal women experience thinning of vaginal mucosa and 657 increased expression of cervical CCR5 receptors (Meditz, Moreau, et al. CROI) that may 658 enhance susceptibility to HIV infection. An approved vaginal microbicide would likely be used 659 in women of all ages; therefore, sponsors are strongly encouraged to collect safety data in this 660 population. Safety, tolerability, and PK data can be collected in a separate trial or by allowing 661 limited enrollment of older women, including postmenopausal women, in phase 3 trials. 662 663 B. Specific Considerations for Effectiveness Trials 664 665 1. Trial Design, Randomization, and Endpoints 666 667 Microbicide effectiveness trials should be randomized, double-blinded, and conducted across 668 multiple sites. HIV seroconversion should be the primary endpoint. Demonstrating reduction in 669 HIV seroconversion requires phase 3 trials with large sample sizes. Phase 2 trials, which do not 670 have sufficient power to show differences in seroconversion, can be conducted to obtain safety 671 data before initiation of large phase 3 trials. For logistical reasons, such as maintaining trial sites 672 and maintaining continued enrollment, phase 2 and phase 3 trials can be combined, with a phase 673 2b lead-in as an initial part of a phase 3 trial. New infections and person-years of exposure 674 obtained in the phase 2 lead-in portion can contribute to the safety and effectiveness of the phase 675 3 portion. Safety and tolerability from the first several hundred women enrolled in the phase 2b 676 lead-in should be evaluated before accrual in the phase 3 portion is initiated. If seroconversion 677 endpoints from the phase 2 lead-in portion are unblinded and evaluated before starting the phase 678 3 portion, then statistically appropriate methods should be used for combining the results of the 679 phase 2 and phase 3 data. This design offers the advantages of enrolling fewer new subjects than 680 if the phase 2 and phase 3 trials were conducted separately, and of allowing more safety 681 evaluations before expanding the trial enrollment. 682 683 2. Choice of Comparator 684 685 We recommend a vaginal microbicide placebo as the comparator in effectiveness trials when 686 investigational drug products are added to a background of other HIV prevention modalities such 687 as condoms, counseling, and even oral PrEP (as deemed appropriate and acceptable by local 688 jurisdictions; see section III.A.4., Efficacy Considerations). The vehicle (or excipient) 689 component may not serve as an acceptable placebo unless it is known that there are no beneficial 690 or harmful effects resulting from the vehicle. Clinical data support HEC gel as an acceptable 691 placebo because outcomes in the HEC gel arm were shown to be no more than 2.4 percent worse 692 than the condom-only arm (or no-treatment arm) in HIV prevention trial HPTN 035 (Karim, 693 Richardson, et al. 2011). 694 695 Following approval of a vaginal microbicide, an active-controlled noninferiority trial design 696 comparing the candidate microbicide to the approved microbicide could be considered 697 appropriate. A noninferiority design depends on the ability to define the magnitude of the

16 Contains Nonbinding Recommendations Draft — Not for Implementation 698 contribution of the new active control such that a reliable noninferiority margin can be 699 calculated.13 700 701 Designing a trial using oral emtricitabine/tenofovir as a comparator will be challenging. A 702 superiority trial is uncomplicated from a statistical perspective, but will likely need large 703 numbers of subjects to demonstrate a treatment effect in the setting of even moderately good 704 adherence. A noninferiority comparison to oral emtricitabine/tenofovir presents methodological 705 challenges because of uncertainty of the assay sensitivity of oral emtricitabine/tenofovir as an 706 active control. In previous trials of oral emtricitabine/tenofovir the overall preventive treatment 707 effect ranged from 0 to 75 percent depending on subject adherence (Grant, Lama, et al. 2010; 708 Baeten, Donnell, et al. 2012; Thigpen, Kebaabetswe, et al. 2012; Van Damme, Corneli, et al. 709 2012). Given this historical data, defining a noninferiority margin will be problematic. Sponsors 710 considering an active-controlled noninferiority design should discuss protocol proposals with the 711 FDA well in advance of trial initiation. Use of a future-approved vaginal microbicide as an 712 active control could result in similar methodologic challenges as for oral emtricitabine/tenofovir 713 depending on the effectiveness (and associated confidence intervals) of the approved vaginal 714 microbicide in previously conducted trials. 715 716 3. Enrollment Criteria 717 718 Healthy, non-HIV-infected adult women should be enrolled in phase 3 trials. Screening 719 evaluation should include medical history, physical examination including pelvic examination, 720 and the following laboratory tests: 721 722  HIV serology 723  Serum hematology and chemistry profile 724  Urine or serum beta human chorionic gonadotropin 725  Testing for STIs, bacterial vaginosis, and vaginal candidiasis 726  Papanicolaou smear 727 728 At least two negative HIV serology test results are necessary to confirm lack of seroconversion 729 before trial entry. Samples for reverse transcription (or transcriptase) polymerase chain reaction 730 (RT-PCR) testing should be obtained and stored at screening and baseline visits (see section 731 III.C.1., Clinical Virology). 732 733 Subjects diagnosed with treatable STIs should receive appropriate treatment and these subjects 734 can be considered for enrollment following resolution of infection. Subjects should not be 735 pregnant and should be willing to prevent pregnancy for the trial duration. Tampon use should 736 not be considered an exclusion criterion. Collecting safety and PK data when the microbicide is 737 co-administered with a tampon is important because tampon use may be common in certain 738 populations. Including tampon users may provide data applicable to such populations. 739

13 See the draft guidance for industry Non-Inferiority Clinical Trials. When final, this guidance will represent the FDA’s current thinking on this topic.

17 Contains Nonbinding Recommendations Draft — Not for Implementation 740 4. Trial Procedures 741 742 HIV seroconversion data, as measured by an approved HIV antibody assay, should be obtained 743 through monthly testing. Safety evaluations should be conducted at scheduled intervals as 744 discussed previously. Data pertaining to type and duration of hormonal contraception used 745 should be collected as part of concomitant medication assessment to allow analysis (subgroup or 746 sensitivity analyses) for any effects of hormonal contraception on microbicide safety or efficacy. 747 Data related to sexual behavior, coital frequency, condom use, microbicide adherence, and 748 tampon use should be collected through interviews and subject diary cards. Subjects who 749 seroconvert during the course of the trial should be referred for HIV treatment according to the 750 local standard of care. 751 752 5. Trial Duration 753 754 All enrolled subjects should be followed for a minimum period of 12 months and should be 755 followed until the last enrolled subject completes the trial and at least 50 percent of the subjects 756 have received 24 months follow-up. Data reflecting safety over a dosing duration of at least 12 757 months is important because an approved microbicide may be used indefinitely and new safety 758 concerns from chronic exposure may arise. Longer duration data also captures product 759 adherence, which may reduce over time with diminishing effectiveness as observed in some 760 biomedical prevention trials (Karim, Karim, et al. 2010; Grant, Lama, et al. 2010). We do not 761 recommend large trials of short duration because they are not as informative with respect to 762 longer term use. 763 764 If interim analyses suggest the targeted number of HIV seroconversions may be reached in 765 advance of the recommended duration of follow-up, sponsors are advised to engage in 766 discussions with the FDA before making decisions about trial termination. Before trial 767 termination, sponsors should provide the FDA with exposure estimates that include the number 768 of subjects with 12, 18, and 24 months follow-up at the time of intended trial termination. 769 Estimations should account for expected lost-to-follow-up and other anticipated reasons for 770 subject discontinuation. 771 772 6. Statistical Considerations 773 774 a. Endpoint analysis 775 776 The primary endpoint should be HIV seroconversion rate per person-year of drug product use. 777 Drug product use should be calculated as the time from provision to the subject of the 778 investigational drug product or placebo until the time when the subject has completed or been 779 discontinued from the trial. Drug product use should not be adjusted for actual use or 780 compliance. Calculations based on person-years of drug product use are preferred to the 781 absolute seroconversion rate because the former accounts for differential dropout rates between 782 arms. In the event of markedly different dropout rates between arms, sensitivity analyses should 783 be provided. Cox proportional hazards regression or Poisson model analysis generally are 784 accepted methods of analysis. Cox proportional hazards regression also can be used with

18 Contains Nonbinding Recommendations Draft — Not for Implementation 785 adjustment for protocol-specified baseline covariates. Survival curves and hazard curves should 786 be plotted to ensure absence of convergence or crossover in hazard rates. 787 788 b. Strength of evidence 789 790 Drug product approval should be supported by evidence from at least two independent trials, 791 each convincing on its own.14 A statistically significant treatment effect compared to control for 792 a superiority trial is a two-sided p-value less than 0.05. Trials with strong internal consistency 793 increase confidence whereas lack of consistency reduces confidence in the result. Conducting 794 two independent trials simultaneously (or nearly so) may be feasible and could avoid potential 795 ethical difficulties if one trial finishes first and shows a statistical significant treatment effect. 796 797 Data from a single large phase 3 trial also may be acceptable. Formally, two independent trials 798 each statistically significant at the two-sided 0.05 level provide a strength of evidence equivalent 799 to a single trial statistically significant at the two-sided 0.001 level. However, accepting data 800 from a single large trial is contingent on several variables, including the generalizability of 801 results to a broader population and internal consistency across subgroups and sites within the 802 trial. Sponsors are advised to refer to FDA guidance for details.15 803 804 Large sample sizes and other challenges with conducting HIV prevention trials may provide an 805 incentive for microbicide sponsors to combine efforts toward collaborative trials. A 806 collaborative trial design evaluating multiple investigational drug products offers the advantage 807 of a single control arm serving as comparator for each drug product. Increasing the control arm 808 size increases the power to show effect for the individual investigational drug product. This 809 design also permits the evaluation of combination products. Safety and effectiveness 810 comparisons between microbicides can provide additional useful information. 811 812 c. Adherence 813 814 Male condoms (or other proven prevention interventions) are expected to reduce HIV 815 transmission; therefore, HIV infection is likely to occur more frequently in subjects not using 816 condoms (or another intervention) irrespective of the trial arm. A larger treatment difference is 817 expected among subjects compliant with the microbicide but not compliant with condoms in case 818 of an efficacious drug product. Secondary analyses in this subgroup, defined based on post- 819 randomization assessments of drug product adherence, may provide supportive evidence of 820 efficacy. Under no circumstances would a compliance-based secondary analysis demonstrating 821 efficacy be considered as a refutation of primary intent-to-treat analysis that failed to 822 demonstrate efficacy. 823 824 We do not recommend drug product adherence incentives, other than coaching and counseling, 825 because incentives will not be available after the drug product is on the market. A trial using 826 them could exaggerate potential benefit. Incentives that encourage subjects to return for study

14 See the guidance for industry Providing Clinical Evidence of Effectiveness for Human Drug and Biological Products.

15 Ibid

19 Contains Nonbinding Recommendations Draft — Not for Implementation 827 visits or maintain contact with site staff are acceptable because these ensure data collection in 828 subjects both on and off their assigned treatments. 829 830 d. Interim analysis and data monitoring committee 831 832 The plan for interim analyses to assess futility and safety should be finalized before trial 833 initiation, and included in the statistical analysis plan. Based on interim findings, a trial may be 834 terminated early for futility if the conditional power is low. Interim findings such as rate of 835 condom usage or specific local practices affecting HIV transmission rate should guide sample 836 size adjustments in an ongoing trial. Such increases in sample size also should be made in 837 accordance with accepted guidelines for adaptive trial design as documented in the published 838 statistical literature on sample size changes. Safety concerns, including a greater number of HIV 839 seroconversions in the investigational arm, should influence considerations for continuing 840 enrollment or halting the trial. Interim analysis results should be reviewed by an independent 841 data monitoring committee to avoid an impact on trial conduct or recruitment. A detailed charter 842 with the composition of the committee members and the operational details should be provided 843 for FDA review.16 Sponsors should remain blinded to individual subject and investigator data. 844 845 e. Missing data and sensitivity analysis 846 847 The sponsor should minimize lost-to-follow-ups and other types of missing data with appropriate 848 planning, including use of incentives that do not interfere with the clinical trial interpretations. 849 850 In addition to the analysis mentioned in section III.B.6.a., Endpoint analysis, sponsors should 851 perform sensitivity analysis where all lost-to-follow-ups are imputed by the placebo arm hazard. 852 853 7. Combination Products 854 855 This section discusses combination products that include two or more microbicide drug products 856 or microbicide-device products. Sponsors are encouraged to refer to other FDA guidance for 857 developing combination products.17 858 859 In general, the following information evaluating the combination microbicide is needed before 860 efficacy studies of a combination product: cell culture combination activity data, nonclinical 861 toxicity for each drug, human safety data from clinical trials for each drug, information 862 supporting selection of proposed doses, and drug-drug interaction data (if applicable). If 863 overlapping toxicity is observed in nonclinical studies, then nonclinical toxicity studies with the 864 combination product may be necessary. Trial designs should include provisions for 865 demonstrating the contribution of each component to the desired effect. Establishing the 866 contribution of each component generally can be accomplished using factorial designs or

16 See the guidance for clinical trial sponsors Establishment and Operation of Clinical Trial Data Monitoring Committees.

17 See the guidance for industry Nonclinical Safety Evaluation of Drug or Biologic Combinations and the draft guidance for industry Codevelopment of Two or More Unmarketed Investigational Drugs for Use in Combination. When final, this guidance will represent the FDA’s current thinking on this topic.

20 Contains Nonbinding Recommendations Draft — Not for Implementation 867 modified factorial designs. Because the development pathway depends on the specific drugs 868 combined, the approach may vary for different types of combination products. We encourage 869 sponsors to discuss their specific combination product early in the development program. 870 871 If one of the components of the combination is marketed as the same vaginal formulation and 872 dose, regulatory requirements may not follow the paradigm outlined above; therefore, sponsors 873 are advised to seek FDA input early in the development process. 874 875 Additional considerations for microbicide-device combinations and combinations for multiple 876 indications are discussed below. 877 878  Vaginal microbicide plus device 879 880 Examples of microbicide-device combination include a microbicide combined with a 881 condom, cervical cap, or diaphragm. A microbicide-device combination will be reviewed 882 by both CDER and CDRH. The FDA Office of Combination Products (OCP) determines 883 which Center is assigned the lead review, and which Center provides a consultative or 884 collaborative review. Both Centers work together through the review process by 885 providing expertise for individual components of the combination. The OCP has a 886 formal product jurisdiction process that can be initiated by submitting a Request for 887 Designation.18 888 889 As previously mentioned in section III.A.1.c., Other nonclinical studies, sponsors should 890 conduct studies evaluating effects of the microbicide on the integrity and function of the 891 device. Stability data for the combination product will be needed if the microbicide 892 formulation used in the microbicide-device combination is different from the stand-alone 893 microbicide. 894 895  Combination product intended for multiple indications 896 897 Developing combination products where each constituent is intended for a different 898 function or indication (e.g., HIV prevention plus contraception) involves complex 899 regulatory considerations. A key consideration for such a multi-indication combination 900 product is whether individual components are investigational or marketed for the 901 respective indication being pursued. If a component is an unmarketed investigational 902 drug, consideration should be given to whether the drug is approved as an oral 903 formulation or as another vaginally administered formulation for the indication being 904 pursued. We anticipate that a combination product seeking multiple indications will need 905 input from multiple CDER divisions each providing regulatory expertise for the specific 906 indication. 907 908 The development pathway for multi-indication combination products involves unique and 909 complex regulatory challenges. Because regulatory advice depends on combination 910 product characteristics and consultations within CDER, sponsors are encouraged to

18 See the guidance for industry How to Write a Request for Designation (RFD).

21 Contains Nonbinding Recommendations Draft — Not for Implementation 911 approach the FDA with questions for their specific combination product and the type of 912 indications they plan to pursue. 913 914 8. Risk-Benefit Considerations 915 916 For the overall risk-benefit assessment, the totality of data should be considered. Benefit in HIV 917 prevention trials should be measured as the percent reduction in HIV transmission during the 918 trial period. Microbicide effectiveness trials should be powered to demonstrate at least a 33 919 percent reduction in HIV seroconversion. Lower rates of reduction in a clinical trial may not 920 translate into a clinically relevant effect because lower rates of adherence or other factors with 921 real-world use may further reduce the actual risk reduction. However, we recognize that lower 922 reductions may have an effect on transmission rates in high HIV prevalence areas. The percent 923 reduction in HIV and the toxicity profile of a microbicide are critical in the decision-making 924 process. One concern in deciding risk-benefit is the potential for behavioral disinhibition in the 925 real-world setting, including condom migration (decrease in condom usage secondary to 926 increased microbicide uptake), that can adversely affect transmission rates. Behavioral data, 927 rates of other STIs, and frequency of self-reported condom use collected in trials include the 928 types of data important for evaluation for the final assessment. Resistance development may be 929 an additional concern, particularly for a systemically absorbed antiretroviral drug product. 930 931 C. Other Considerations 932 933 1. Clinical Virology 934 935 Sensitivity and specificity of all assays used to verify HIV status should be validated for a panel 936 of viral isolates representing the most common circulating strains endemic to the trial region. 937 Sponsors should collect and store baseline samples for RT-PCR for all subjects. Testing of 938 stored baseline/screening samples with RT-PCR should be performed to confirm the infection 939 status of subjects who seroconvert during the trial. Verification by RT-PCR is necessary because 940 the antibody assays typically used for baseline analyses (e.g., assay for anti-HIV antibodies) lack 941 the sensitivity of the RT-PCR assay for detecting acute infection. HIV-positive subjects missed 942 by the screening HIV assay should not be considered new seroconversions and should be 943 excluded from the primary analysis. 944 945 For a microbicide containing an antiretroviral drug, viral isolates from subjects who seroconvert 946 should undergo genotypic testing. Microbicide failure in these subjects may represent selection 947 and enrichment of resistant HIV-1 variants. This selection could occur during local replication 948 within the vaginal epithelium or during systemic replication, even if there is poor bioavailability. 949 When possible, isolates from a subject’s partner also should be genotypically characterized to 950 assess the potential for transmission of resistant virus. If analysis of HIV-1 isolates identifies 951 novel substitutions not previously analyzed during nonclinical resistance studies, the 952 substitutions should be phenotypically characterized. 953

22 Contains Nonbinding Recommendations Draft — Not for Implementation 954 2. Additional Clinical Pharmacology 955 956 In vivo drug interaction studies should be considered with the candidate microbicide and 957 commonly used vaginal drug products, including the contraceptive ring (e.g., Nuvaring) and 958 commonly used antimicrobials (e.g., metronidazole vaginal suppository or gel), to ensure local 959 release characteristics and systemic exposure, where applicable, are not adversely affected by co- 960 administration. The drug interaction data collected should be applicable to the U.S. population 961 and U.S. medical practice. Preferably, all relevant drug interaction studies would be completed 962 before conduct of large phase 2b or phase 3 trials to allow use of these drug products in the trials. 963 964 Large effectiveness trials should incorporate a PK substudy to allow characterization of local 965 and, if applicable, systemic microbicide exposure in women during actual conditions of use. In 966 addition, PK samples should be obtained from all women at study visits at which HIV testing is 967 performed and the samples archived for future analysis. Because microbicides can be developed 968 with various coitally dependent and coitally independent dosing schemes, the sampling schedule 969 should be discussed with the FDA. Time of previous doses and the time of the sample collection 970 should be recorded for all PK samples. In the event of seroconversion, all samples for the 971 seropositive subjects should be analyzed and compared to all samples for a matched seronegative 972 cohort. PK seroconversion data obtained from a large clinical trial may be useful to evaluate 973 dose, for assessing adherence patterns, and for further analysis in the event of a failed trial. 974 975 3. Chemistry, Manufacturing, and Controls 976 977 Sponsors and applicants should refer to FDA and ICH quality guidances for industry for general 978 chemistry, manufacturing, and controls (CMC) recommendations for drug substance (active 979 pharmaceutical ingredient) and drug product (finished dosage form). Drug product manufacture 980 must be in compliance with current good manufacturing practice requirements (21 CFR parts 981 210 and 211). CMC considerations specific to vaginal microbicides are discussed below. 982 983 Sponsors should conduct formulation studies to evaluate drug product attributes that affect 984 microbicide quality and performance. Attributes that may affect drug product retention and 985 distribution in the vagina or have effects on vaginal epithelium or influence growth of pathogens 986 including Candida albicans and Escherichia coli should be evaluated (e.g., drug product 987 solubility at different pH environments, drug product stability, rheological characteristics, 988 adhesion of the vehicle). Biocompatibility studies should be conducted for vaginal ring 989 microbicides. Attributes that may affect release rate of the active ingredient such as solubility, 990 solid-state form, or particle size of drug substance should be evaluated. For drug products that 991 support microbial growth, preservative effectiveness studies should be conducted and the drug 992 product formulated with a minimum effective content of preservative. Sponsors should evaluate 993 the safety, suitability, and performance of the proposed container closure system and delivery 994 system (e.g., vaginal applicator). 995 996 Drug product specification should include tests for universal attributes such as identity, strength, 997 and impurities. Depending on the specific dosage form, additional tests should be included in 998 the drug product specification (e.g., viscosity, pH, particle size distribution, homogeneity 999 assurance, dissolution or melting rate, in vitro release rate for controlled-release drug products,

23 Contains Nonbinding Recommendations Draft — Not for Implementation 1000 tensile strength for vaginal rings). Testing for microbial limits also should be part of the drug 1001 product specification. If the active ingredient is a polymer, a test should be included for 1002 molecular weight distribution (e.g., size-exclusion chromatography). If the drug product 1003 contains a preservative, a test for preservative content should be included. 1004 1005 In vitro release testing has been shown effective in assessing the drug product quality and 1006 performance over time for certain semisolid dosage forms. However, sponsors should not use in 1007 vitro testing as a surrogate for in vivo bioavailability or bioequivalence. Further, in vitro release 1008 testing is not considered an appropriate measure for drawing comparisons between different 1009 semisolid formulations or in comparison of similar formulations across different manufacturers. 1010 Sponsors are reminded that the formulation evaluated in phase 3 trials should be identical to the 1011 formulation proposed for marketing. 1012 1013 The stability profile for the drug product should be established under long-term and accelerated 1014 storage conditions using analytical methods capable of detecting physical changes and chemical 1015 degradation. The microbicide should remain stable at a wide range of pH including normal 1016 vaginal pH. Data from initial stability studies should be provided to support the dosing duration 1017 of the proposed clinical trials. Stability data supporting a proposed expiration date for the 1018 commercial drug product are recommended at the time of NDA submission. 1019 1020 4. Condom/Device Function Studies 1021 1022 Condom function studies (clinical trials evaluating effects of a microbicide on the failure rate of 1023 condoms) may be needed for certain microbicides. As mentioned previously in section 1024 III.A.1.c., Other nonclinical studies, results from nonclinical compatibility tests will guide the 1025 need to conduct clinical trials in this area. Sponsors are strongly encouraged to seek input from 1026 the FDA regarding the need for condom function studies and the proposed study design. If a 1027 clinical trial is needed, sponsors are advised to plan trials based on appropriate methodology 1028 (Taylor 2009). 1029 1030 5. Labeling Considerations 1031 1032 The label should emphasize that the drug product is intended for vaginal use only, and that 1033 efficacy with oral and rectal use are not established. The label should also convey that safety 1034 following oral exposure is not known, unless data are available. For vaginal ring drug products, 1035 a label comprehension study may be needed to ensure instructions for intravaginal ring use are 1036 appropriate for the U.S. population. These data should be available for FDA approval. For 1037 vaginal gel drug products using an applicator, similar studies evaluating end-user ability to 1038 correctly apply the drug product should be conducted before approval. 1039

24 Contains Nonbinding Recommendations Draft — Not for Implementation 1040 REFERENCES 1041 1042 ASTM Standard D7661, 2010, Standard Test Method for Determining Compatibility of Personal 1043 Lubricants with Natural Rubber Latex Condoms, ASTM International, West Conshohocken, PA, 1044 DOI: 10.1520/D7661-10 (www.astm.org). 1045 1046 Aziz M and Smith KY, 2011, Challenges and Successes in Linking HIV-Infected Women to 1047 Care in the United States, Clin Infect Dis, 52:S231-237. 1048 1049 Baeten JM, Donnell D, Ndase P et al., 2012, Antiretroviral Prophylaxis for HIV Prevention in 1050 Heterosexual Men and Women, N Engl J Med, DOI: 10.1056/NEJMoa1108524. 1051 1052 Berges BK, Akkina SR, Folkvord JM et al., 2008, Mucosal Transmission of R5 and X4 Tropic 1053 HIV-1 Via Vaginal and Rectal Routes in Humanized Rag2-/- gammac -/- (RAG-hu) Mice, 1054 Virology, 373(2):342-51. 1055 1056 Centers for Disease Control and Prevention (CDC), 2010, HIV/AIDS Fact Sheet for HIV in the 1057 United States in Women, February. 1058 1059 Denton PW, Estes JD, Sun Z et al., 2008, Antiretroviral Pre-Exposure Prophylaxis Prevents 1060 Vaginal Transmission of HIV-1 in Humanized BLT Mice, PLoS Med, 5(1):e16. 1061 1062 Fuchs EJ, Lee LA, Torbenson MS et al., 2007, Hyperosmolar Sexual Lubricant Causes Epithelial 1063 Damage in the Distal Colon: Potential Implication for HIV Transmission, J Infect Dis, 1064 195(5):703-10. 1065 1066 Grant RM, Lama JR, Anderson PL et al., 2010, Pre-Exposure Chemoprophylaxis for HIV 1067 Prevention in Men Who Had Sex With Men, N Engl J Med, 363:2587-99. 1068 1069 Hall HI, Song R, Rhodes P et al., 2008, Estimation of HIV Incidence in United States, JAMA, 1070 300(5):520-529. 1071 1072 Karim QA, Karim SA, Frohlich JA et al., 2010, Effectiveness and Safety of Tenofovir Gel, An 1073 Antiretroviral Microbicide, for the Prevention of HIV Infection in Women, 1074 Sciencexpress.org/19, July. 1075 1076 Karim QA, Richardson BA, Ramjee G et al., 2011, Safety and Effectiveness of BufferGel and 1077 0.5% PRO2000 gel for the Prevention of HIV Infection in Women, AIDS, 25(7):957-66. 1078 1079 Kulczycki A, Kim Dhong-Jin, Duerr A et al., 2004, The Acceptability of the Female and Male 1080 Condom: A Randomized Crossover Trial, Perspectives on Sexual and Reproductive Health, 1081 36(3):114-119. 1082 1083 Lagakos SW, Gale AR, Editors, 2008, Methodological Challenges in Biomedical HIV 1084 Prevention Trials, Committee on the Methodological Challenges in HIV Prevention Trials, 1085 Institute of Medicine of National Academies, The National Academies Press Washington D.C. 1086

25 Contains Nonbinding Recommendations Draft — Not for Implementation 1087 Meditz A, Moreau K, Gozansky et al., Healthy Post-Menopausal Women Have Higher 1088 Percentages of CCR5 Cervical CD4 T Cells Compared to Pre-Menopausal Women: Implications 1089 for HIV Transmission, 18th Conference on Retroviruses and Opportunistic Infections Oral 1090 Abstract. 1091 1092 Moss GB, Celemetson D, D’Costa L et al., 1991, Association of Cervical Ectopy With 1093 Heterosexual Transmission of Human Immunodeficiency Virus: Results of a Study of Couples 1094 in Nairobi, Kenya, Journal of Infect Dis, 164:588-91. 1095 1096 Myer L, Kuhn L, Stein ZA et al., 2005, Intravaginal Practices, Bacterial Vaginosis, and 1097 Women’s Susceptibility to HIV Infection: Epidemiological Evidence and Biological 1098 Mechanisms, Lancet Infect Dis, 5(12):786-794. 1099 1100 Nelson RM, Lewis LL, Struble K, Wood SF, 2010, Ethical and Regulatory Considerations for 1101 the Inclusion of Adolescents in HIV Biomedical Prevention Research, J Acquir Immune Defic 1102 Syndr, 54:S18-21. 1103 1104 Poles MA, Elliott J, Taing P et al., 2001, A Preponderance of CCR5(+) CXCR4(+) Mononuclear 1105 Cells Enhances Gastrointestinal Mucosal Susceptibility to Human Immunodeficiency Virus Type 1106 1 Infection, J Virol, 75:8390-9. 1107 1108 Taylor DJ, 2009, Issues in the Design, Analysis, and Interpretation of Condom Functionality 1109 Studies, Contraception, 80:237-244. 1110 1111 Thigpen MC, Kebaabetswe PM, Paxton LA, et al, 2012, Antiretroviral Preexposure Prophylaxis 1112 for Heterosexual HIV Transmission in Botswana, N Engl J Med, DOI: 1113 10.1056/NEJMoa1110711. 1114 1115 UNAIDS, 2011, Report on the Global HIV/AIDS Epidemic. 1116 1117 Van Damme L, Corneli A, Ahmed K et al., 2012, Preexposure Prophylaxis for HIV Infection 1118 Among African Women, N Engl J Med, DOI: 10.1056/NEJMoa1202614. 1119 1120 Wu L, Kewalramani V, 2006, Dendritic Cell Interactions With HIV: Infection and Viral 1121 Dissemination, Nat Rev Immunol, 6(11):859-868. 1122

26